Image Reconstruction without an Inertial Navigation System using Backprojection Autofocus for Synthetic Aperture Radar

We propose an extension of the backprojection autofocus approach to reconstruct synthetic aperture radar images with high quality from extreme coarsely measured flight paths. Usually, expensive and heavy inertial navigation systems are used to measure flight paths with high precision. Small antenna position errors up to approximately one half of the range resolution can be corrected by state-of-the-art autofocus techniques. Our approach is able to correct position errors much larger than this upper bound. For this purpose, we use the idea of the backprojection autofocus by Ash and estimate pulse-by-pulse phase errors by maximizing image sharpness. In contrast to Ash, we use additionally the estimated phase error of each considered pulse to update the aperture positions of all subsequent pulses. The evaluation of our autofocus technique using real airborne X-band data acquired in spotmode shows that the proposed method can correct antenna position errors of more than 12 m. Thus, our algorithm allows the usage of much cheaper or even no inertial navigation systems.